Light emitting diodes (“LEDs”) are p-n junction devices that have been found to be useful in various roles as the field of optoelectronics has grown and expanded over the years. Devices that emit in the visible portion of the electromagnetic spectrum have been used as simple status indicators, dynamic power level bar graphs, and alphanumeric displays in many applications, such as audio systems, automobiles, household electronics, and computer systems, among many others. Infrared devices have been used in conjunction with spectrally matched phototransistors in optoisolators, handheld remote controllers, and interruptive, reflective, and fiber-optic sensing applications.
An LED operates based on the recombination of electrons and holes in a semiconductor. When an electron carrier in the conduction band combines with a hole in the valence band, it loses energy equal to the bandgap in the form of an emitted photon; i.e., light. The number of recombination events under equilibrium conditions is insufficient for practical applications but can be enhanced by increasing the minority carrier density.
In an LED, the minority carrier density is conventionally increased by forward biasing the diode. The injected minority carriers radiatively recombine with the majority carriers within a few diffusion lengths of the junction edge. Each recombination event produces electromagnetic radiation, i.e., a photon. Because the energy loss is related to the bandgap of the semiconductor material, the bandgap characteristics of the LED material have been recognized as being important.
As with other electronic devices, however, there exists both the desire and the need for more efficient LEDs, and in particular, LEDs that will operate at higher intensity while using less power. Higher intensity LEDs, for example, are particularly useful for displays or status indicators in various high ambient environments. There also is a relation between intensity output of the LED and the power required to drive the LED. Low power LEDs, for example, are particularly useful in various portable electronic equipment applications. An example of an attempt to meet this need for higher intensity, lower power, and more efficient LEDs may be seen with the development of the AlGaAs LED technology for LEDs in the red portions of the visible spectrum. A similar continual need has been felt for LEDs that will emit in the green, blue and ultraviolet regions of the visible spectrum. For example, because blue is a primary color, its presence is either desired or even necessary to produce full color displays or pure white light.
The common assignee of the present patent application was the first in this field to successfully develop commercially viable LEDs available in large quantities and that emitted light in the blue color spectrum. These LEDs were formed in silicon carbide; and examples are described in U.S. Pat. Nos. 4,918,497 and 5,027,168 to Edmond each titled “Blue Light Emitting Diode Formed in Silicon Carbide.”
Other examples of such a blue LED are described in U.S. Pat. No. 5,306,662 to Nakamura et al. titled “Method of Manufacturing P-Type Compound Semiconductor” and U.S. Pat. No. 5,290,393 to Nakamura titled “Crystal Growth Method for Gallium Nitride-Based Compound Semiconductor.” U.S. Pat. No. 5,273,933 to Hatano et al. titled “Vapor Phase Growth Method of Forming Film in Process of Manufacturing Semiconductor Device” also describes LEDs formed of GaInAlN on SiC substrates and Zinc Selenide (ZnSe) on gallium arsenide (GaAs) substrates.
General discussions of LED technology can be found at Dorf, The Electrical Engineering Handbook, 2d Ed. (1997, CRC Press), at pages 1915-1925, section 83.1, “Light Emitting Diodes,” and in Sze, Physics of Semiconductor Devices, at pages 681 ff, Chapter 12, “LED and Semiconductor Lasers” (1981, John Wiley & Sons, Inc.).
As known to those familiar with photonic devices such as LEDs, the frequency of electromagnetic radiation (i.e., the photons) that can be produced by a given semiconductor material are a function of the material's bandgap. Smaller bandgaps produce lower energy, longer wavelength photons, while wider bandgap materials are required to produce higher energy, shorter wavelength photons. For example, one semiconductor commonly used for lasers is indium gallium aluminum phosphide (InGaAlP). Because of this material's bandgap (actually a range of bandgaps depending upon the mole or atomic fraction of each element present), the light that InGaAlP can produce is limited to the red portion of the visible spectrum, i.e., about 600 to 700 nanometers (nm).
Working backwards, in order to produce photons that have wavelengths in the blue or ultraviolet portions of the spectrum, semiconductor materials are required that have relatively large bandgaps. Typical candidate materials include silicon carbide (SiC) and Group III nitrides, particularly gallium nitride (GaN), and ternary and tertiary Group III nitrides such as aluminum gallium nitride (AlGaN), indium gallium nitride (InGaN), and aluminum indium gallium nitride (AlInGaN).
Shorter wavelength LEDs offer a number of advantages in addition to color. In particular, when used in optical storage and memory devices (e.g., “CD-ROM” or “optical disks”), their shorter wavelengths enable such storage devices to hold proportionally more information. For example, an optical device storing information using blue light can hold approximately 4 times as much information as one using red light, in the same space.
The Group III nitrides, however, are attractive LED candidate material for green, blue, and UV frequencies because of their relatively high bandgaps (i.e., GaN is 3.36 eV at room temperature) and because they are direct bandgap materials rather than an indirect bandgap material. As known to those familiar with semiconductor characteristics, a direct bandgap material is one in which an electron's transition from the valence band to the conduction band does not require a change in crystal momentum for the electron. In indirect semiconductors, the alternative situation exists; i.e., a change of crystal momentum is required for an electron's transition between the valence and conduction bands. Silicon and silicon carbide are examples of such indirect semiconductors.
Generally speaking, LEDs formed in direct bandgap materials will perform more efficiently than ones formed in indirect bandgap materials because the photon from the direct transition retains more energy than one from an indirect transition.
The Group III nitrides suffer from a different disadvantage, however: the failure to date of any workable technique for producing bulk single crystals of Group III nitrides which could form appropriate substrates for Group III nitride photonic devices. As is known to those familiar with semiconductor devices, they all require some sort of structural substrate. Typically, a substrate formed of the same materials as the active region of a device offers significant advantages, particularly in crystal growth and matching. Because Group III nitrides have yet to be formed in such bulk crystals, however, Group III nitride photonic devices must be formed in epitaxial layers on different—i.e., other than Group III nitride—substrates.
Using different substrates, however, causes an additional set of problems, mostly in the areas of crystal lattice matching and thermal coefficients of expansion (TCEs). In almost all cases, different materials have different crystal lattice parameters and TCEs. As a result, when Group III nitride epitaxial layers are grown on a different substrate, some crystal mismatch will occur, and the resulting epitaxial layer is referred to as being either “strained” or “compressed” by these mismatches. Such mismatches, and the strain they produce, carry with them the potential for crystal defects which in turn affect the electronic characteristics of the crystals and the junctions, and this correspondingly tend to degrade or even prevent the performance of the photonic device. Such defects are even more problematic in higher power structures.
To date, a common substrate for Group III nitride devices has been sapphire; i.e., aluminum oxide (Al2O3). Sapphire is optically transparent in the visible and UV ranges, but is unfortunately insulating rather than conductive, and carries a lattice mismatch with (for example) gallium nitride of about 16%. In the absence of a conductive substrate, “vertical” devices (those with contacts on opposite sides) cannot be formed, thus complicating the manufacture and use of the devices.
As a particular disadvantage, horizontal structures (those with contacts on the same side of the device), such as those required when Group III nitride layers are formed on sapphire, also produce a horizontal flow of current and therefore the current density through the layer is substantially increased. This horizontal current flow puts an additional strain on the already-strained (e.g., a 16% lattice mismatch between GaN and sapphire) structure and accelerates the degradation of the junction and the device as a whole.
Gallium nitride also carries a lattice mismatch of about 2.4% with aluminum nitride (AlN) and a 3.5% mismatch with silicon carbide. Silicon carbide has a somewhat lesser mismatch (only about 1%) with aluminum nitride.